Browsing the contents page, you’ll soon notice that supporting massive network traffic growth is one of the major themes for issue #6. And this includes a fascinating interview with Peter Winzer of Nokia Bell Labs, which points to enormous value for photonic integrated circuits (PICs) over the coming decade as ‘engines’ for parallel solutions.

The discussion continues with contributions from Macom and Oclaro exploring options for cloud data centres and ways to ‘move the needle on further on available bandwidth’. But that's not the only talking point you'll find in issue #6.

We’ve reached out to Nick Psaila, CEO and co-founder of Optoscribe, for an update on PIC manufacturing, and we hear from Scott Jordan, head of photonics market segment at Physik Instrumente (PI), on how advances in parallel nanoalignment can benefit device production.  

Imec shares developments in chip and systems technologies, specifically its silicon photonics-based platform for high-speed optical links for data, telecom and sensing applications. And, pushing further into emerging markets, Sascha Geidel and Thomas Otto from Fraunhofer ENAS explore what it takes to configure photonic integrated circuits for use in biological sensing or environmental analysis. 

Finally, rounding off issue #6, EPIC reports on pilot line opportunities for Mid-IR sensing and we reveal why there’s even more to look forward to at PIC International 2018.

PIC International 2018 will take place 10-11 April at the Sheraton Airport Hotel, Brussels. Secure your place by registering your interest today - visit  https://picinternational.net/register.

Nominations for the PIC Awards 2018 are open until 17 November 2017 – to see the full list of categories, visit http://www.picawards.net/categories

Enjoy the issue!

James

Email - james.tyrrell@angelbc.com

To handle this huge volume of data traffic, Cloud Data Centre operators such as Amazon, Microsoft, Google and Facebook are rapidly building out their data centres and simultaneously boosting connectivity by extending data delivery speeds to 100G and 400G and beyond.  Space savings, power consumption and cost efficiencies are the critical considerations guiding data centre operators’ infrastructure build-out strategies as they seek to lower their average per-bit delivery costs. 

It is therefore imperative that they maximize the data throughput per faceplate in their data centre switches at the lowest possible cost, resulting in maximized bandwidth density in Cloud Data Centres, it is essential to improve the integration of analog and photonic content within the module, be it a CFP, CFP2, CFP4, QSFP28, OSFP or QSFP-DD form factor. 

In parallel, as is the case with network service providers, Cloud Data Centres need to maximize the data throughput over their installed fibre in a cost efficient manner. Advanced modulation schemes that are efficient to implement in Silicon such as PAM4 increase the amount of data transmitted over each existing fibre connection while reducing the cost per bit significantly, hence the importance of PAM4 as a key technology in next generation data centres.

Individual Cloud Data Centres also need to be connected to each other at high bandwidth speeds to meet resiliency and data redundancy requirements, spanning distances up to 100km. Data centre interconnects (DCIs) are therefore rapidly transitioning to 100G and 400G, driving a tremendous demand for high-speed metro optical links.

As of fiscal year 2016 MACOM enabled over a million 100G modules going into data centre and enterprise applications. What’s even more exciting is that we believe this is just the beginning of a ramp that will see demand for 100G connectivity within data centres triple over the next four years.

Building on our success with EFT lasers, MACOM has moved quickly to exploit EFT to enable the seamless integration of optical devices spanning lasers, modulators and multiplexers onto a single chip, yielding the industry’s first silicon photonic integrated circuit (PIC) integrated with lasers (L-PIC™) for 100G. By solving the key challenge of aligning lasers to the silicon PIC with high yield and high coupling efficiency, MACOM is making the adoption of silicon PICs a reality for high speed, high density optical interconnects in Cloud Data Centres.

Flagship products

With the recent definitive agreement to acquire AppliedMicro, MACOM plans to solidify its technology leadership by marrying APM’s PAM4 100G Single-lambda PHY technology that increases bandwidth density by 4X with MACOM’s L-PIC, and combined with the PHY will enable a highly streamlined 400G transceiver that will quadruple the data throughput over existing fibre infrastructure.  The completion of the portfolio on the PAM4, analog and photonic side positions us ideally for PAM4 adoption not just at 100G, but more importantly at 400G. Within the data centres, there are two standards – to date everything has been NRZ, which uses MACOM 25G lasers, 100G laser drivers, VCSEL drivers and CDRs. The second is PAM4 modulation, a key element moving to 400G. MACOM estimates more than half of the industry will be moving forward with PAM4 within Data Centres at 100G. The future integration of AppliedMicro and MACOM will combine our analog and photonic content for the complete front end capability, combined with the DSP in the PHY with SerDes as well as data converters. 

The wisdom of this approach was recently ratified by the IEEE, who announced standardization around the proposal which was championed by AppliedMicro and Cisco to implement single lambda PAM4 as the industry standard again for both 100G and 400G. 

The continued strategic expansion of MACOM’s product portfolio for Cloud Data Centres will encompass key protocols and span all formats, giving MACOM the agility to respond quickly to evolving market dynamics. Our flagship products are technology agnostic, and can be readily deployed to meet the unique needs of our tier-one customers. 

MACOM’s competencies in Cloud Data Centre connectivity culminate in our ability to provide customers with the entire optical subassembly, leveraging our portfolio of transmit optical subassemblies (TOSAs) and receive optical subassemblies (ROSAs) to optimize and validate chip-level semiconductor integration and packaging. For customers who may not have in-house optical packaging expertise, MACOM can provide the subassembly and/or reference designs that enable them to take advantage of our advanced optical components to speed the development of next generation optical transceivers.

The full portfolio from Switch to Fibre

The overarching goal for me and my team is to stretch the envelope in high-speed and high-capacity optical fibre communications; charting out possibilities for future commercial optical networking hardware. In this vein, we have been continuously holding the world record in single-carrier interface rates over the past 12 years, from 100-Gbit/s on/off keying in 2005 all the way to 1-Tbit/s coherent quadrature phase shift keying in 2015.

In terms of capacity (or rather spectral efficiency, which is the key to capacity from a transponder perspective), we have been pushing systems to record rate-reach trade-offs, using flexibly adapting probabilistically shaped modulation constellations, which we recently field-trialed on an installed trans-Atlantic cable operated by Facebook to show unprecedented scalability of the deployed submarine infrastructure. 

While pushing the limits on the experimental side, we also explored limits from a theoretical angle and realized that fundamental Shannon limits will be soon halting our ability to scale systems in a conventional manner. We have therefore advocated for spatial parallelism as the only way to solve the increasingly critical network scalability problem. As with all parallel solutions (think “multi-core microprocessors”), integration is key to arrive at cost-effective products. Hence, massive arrays of opto-electronic components will play a dominant role over the coming decade, and PICs are the only way to effectively address this problem. You can read more about these needs here.

For many years, network traffic has been scaling at about 60% per year (doubling every 18 months), and there is no reason to believe that this trend is going to change. On the other hand, interface rates have been showing a long-term scaling trend of around 20% per year, and so have fibre capacities over the past decade, (quite unlike the 100% annual capacity scaling they showed in the 1990s) again with no reason to believe that these trends are going to change for the better (they may in fact change for the worse owing to both engineering and fundamental scalability limitations). 

These glaring scaling disparities mean that without massively parallel solutions it will be “game over” in terms of supporting the kind of massive network traffic growth that our society has gotten so accustomed to. In that arena of massive scaling disparities, PICs will play an ever-increasing role as the engine that enables massive parallelism on both client and line interfaces.

Just think of client optics: in order to satisfy the long-term scaling trend of router blades at 40% per year, with the most advanced router blades today offering a commercial 2.4 Tbit/s per blade), we expect router blades supporting several 10 Tbit/s by ca. 2025, which translates into several client interfaces of 10 Tbit/s each! This may sound like science fiction today, but in less than 10 years it is likely to be commercial reality. And no matter how you slice the pie, you won’t be able to commercially implement 10 Tbit/s on a single carrier by then, so massively integrated parallelism in the form of 10 x 1 Tbit/s or 25 x 400 Gbit/s will be the only way -the way of the PIC!

As a first important step, I would like to see massively parallel transmitter and receiver PICs that enable 10-Tbit/s client and line interfaces in research. Once this is demonstrated in research, many practical aspects will still have to be resolved in order to implement corresponding products, including thermal management, opto-electronic packaging, fibre attachment, and optics-electronics co-integration, to name just a few of the challenges. 

If our industry wants to continue the decade-long sustained scaling that our information and communication society has learned to depend on, we need to support 10-Tbit/s class interfaces as commercial products by ca. 2025. Hence, our community in academia and industry alike better hurry up to get there in time!

Recent trends

From a line side transmission perspective, there is a strong initiative to increase baud rates and modulation formats to enable bit rates above 250 Gb/s all the way up to 600 Gb/s. This transition will drive the emergence of new modulation formats that will increase the optical performance that photonic integrated circuits need to achieve. 

One critical requirement that needs to happen is to increase the bandwidth per carrier wavelength as this not only increases the total transmission capacity per fibre, but also helps drive down the cost per bit. A perfect example of this trend was Ciena’s announcement at the most recent OFC Conference. By opening access to its WaveLogic Ai coherent DSP chipset, Ciena is paving the way for optical chip companies to develop programmable modules supporting up to 400 Gb/s with industry-leading baud rates that allow more data to be transported on each wavelength.

Amazingly, people are now even starting to talk about 600 GB/s per wavelength, which will enable the industry to achieve 1.2 TB with two wavelengths from a single module. However, the real Nirvana will be when we can push that bandwidth per wavelength up to 1 TB and beyond. This is where the industry is ultimately heading and it’s going to place a lot of emphasis on the discrete laser and receiver chips as well as more complex photonic integrated circuits.  

Oclaro has many years of experience with integrating different functions with its InP technology enabling it to combine tunable laser designs with Mach Zender Modulation and amplification on a single chip. Benefits include reduced power consumption, higher density solutions (higher aggregate bandwidth on the linecard faceplate), and lower cost per bit while at the same time guaranteeing high field reliability. These integrated solutions have shipped in various module form factors, the latest being the CFP2 ACO for coherent transmission to 250Gbs for long haul, metro and datacentre interconnect applications.

What’s Next?

It is very clear that in the future the ability to design higher levels of integration will be paramount. The continuing thirst for even more bandwidth as we move into the next decade to support the 4Gbit’s to 5Gbit’s market transition in the wireless access market as well as increasing server speeds to 25G and 50G will require companies such as Oclaro to continuously innovate. 

This innovation will encompass all areas of design, tools, process and fabrication technology to drive even higher levels of integration in photonic chips – targeting higher baud rates, increased bit rates, while at the same time shrinking the size of the die and increasing the optical functionality and complexity without increasing the power consumption at the optical subassembly and/or the final module.  

Photonic integration is one of the most important optical technologies of our decade. It represents the fundamental building blocks that the telecom and datcom industry needs to move the needle further in terms of available bandwidth. And I look forward to seeing how far we have come in another year.  With photonic integration, the future is indeed bright!

The photonics industry must focus on what can be done to adopt electronics style assembly processes. In addition, some PIC designs require hermetic packaging, which adds significant expense.  Alternative technologies that avoid fully hermetic sealing should be deployed wherever possible to eliminate this costly element. 

I am very optimistic for the expansion of PIC technologies. I think platforms may achieve a higher market share than perhaps many market reports forecast. The technology has the capability to displace current incumbent, alternative technologies and with enough investment PICs could drive overall product cost down, and meet data centre operators’ challenging cost aspirations. 

With enough volume, silicon photonics can become very cost effective due to the well-established silicon foundry model inherited from the electronics industry, providing it can be packaged cost effectively. To succeed, volume is the key. Once the critical volumes are achieved we will quickly see the cost and yields reach the target levels – historically MEMS followed this same path.  

At the moment, vertical cavity surface emitting lasers (VCSELs) are dominant because they are very low-cost and reliable, but most large data centre operators have a roadmap to single-mode-only architectures which we know is a challenge for VCSELs. If Indium Phosphide can scale in the wafer size and chip complexity, whilst maintaining good yields, there is a chance it could make a dent in data centre client-side interconnects. However, to date the cost has mostly limited this to telecoms and line-side data centre interconnect (DCI) applications.  As it stands however, I believe silicon photonics has a key opportunity to dominate at 400G and beyond within the data centre. There are many in both camps that are fighting to gain market share. 

Fundamentally there has been so much investment and rapid technology development in silicon photonics, that it is very difficult to see it not having a very big part to play in the future of optical interconnects. 

The ability to increase the density of optical channels in a package is a significant benefit. On the PIC side it reduces the chip surface area used for the optical interface, which can help to reduce cost. As the size of the PIC becomes smaller, the need for compact fibre coupling solutions such as Optoscribe’s monolithic glass 3D fibre coupling platform becomes a necessity.  This compact and precise type of coupling platform brings significant benefits and with the move to larger optical channel counts and potentially multicore fibre, which has an inherent density advantage, this smaller footprint will become even more important.

Optoscribe brings other benefits to enable electronics style high volume manufacturing processes for optical transceivers. Products made using these kinds of processes typically use pick and place for ball grid array components and these often need to go through a solder reflow process, which is problematic for many conventional packaging materials such as epoxies due to the high temperature profiles.

I think both Optoscribe and PIC technology have applications in wider telecoms networks as well as datacoms. Starting with the shorter-range side of the network, including potentially 5G mobile infrastructure, there will be a gradual move into longer range market segments as time goes on. On the long haul and ultra-long haul side of the telecoms network there are a number of well-established incumbents using discrete components, which could be hard to displace in the foreseeable future. However, in other segments such as data centre interconnects there is already a precedent of PICs being deployed.  

In addition to this, PICs have a bright future in general purpose computing interconnects. As transceiver functionality transitions from pluggable modules to on-board to co-packaged and ultimately embedded alongside processing electronics, PIC technology will be able to play an increasingly important role in providing high-speed low-latency links as the interconnect requirements for high-speed electronics scales further.

From there, PIC technology will eventually find its way into everyday computing, and eventually right into consumer electronics.  As key industry luminaries have talked about for some time, electrical interconnects have become a particularly significant source of power usage in modern high-speed computers and information processing electronics. Integrated photonics has the opportunity to significantly reduce the power consumption involved in data transmission, including within the next generation of electronic integrated circuits.  

One of the biggest challenges has been building an understanding of the market intelligence to give Optoscribe the independence to be able to take the technology to market. Also, it’s very important to build the right team around the technology, with employees who can transition as the company matures. As the firm evolves from its very early stage, with just a few people, and starts to scale and grow you require a different type of person. 

It is important to realise that this transition needs a specific skill set and that you need to bring in people who have the appropriate experience. This covers all aspects of the business from sales and marketing through to operations and technical management. As the teams grow, the skill sets for managing teams becomes critically important to bringing the company success. Equally important is having the right investors, and finding investors that are willing to support a particular type of business; in our case, innovative hardware technology. We had to find investors that are keen to support a high-tech manufacturing business that involves some level of capital investment in a manufacturing facility and capital expenditure. 

Finding the correct type of premises and location for the business to operate in with access to the relevant staff in the area was important. Fortunately, the optical sector has a good presence in Scotland, which means there is access to a highly educated, skilled workforce, and this has been very beneficial in building the company.

In terms of taking technology from a university environment to a high technology manufacturing model, there was the typical naivety at the early stage where companies have aspirations to address everything under the sun. Then the quick realisation that it is absolutely not a feasible way to build a business and you quickly learn to whittle down to areas in which the technology has the potential to disrupt a growing market. 

One important advantage with the type of technology we offer is to be able to provide a customisation service and offer customers a quick turnaround of prototypes configured precisely around their product design.  To carry out the range of activities here using conventional lithographic technologies would require a substantially larger workforce and facility. The technology and the automated processes we have developed allows us to do a great deal with the current team of 25 employees.

Updated each year to capture the latest industry trends, the conference programme welcomes new speakers and – new for 2018 – includes on-stage industry panels examining high-volume opportunities for PICs, and discussing progress in silicon photonics.  

The PIC awards were a big hit at the 2017 show (see – ‘PIC Awards in pictures’, issue #5) and with two new awards for 2018 there are even more opportunities to join in the celebration. You’ll find full details at www.picawards.net/categories and nominations are open until 17 November 2017.

Book your tickets now

All of the speaker details can be found on the conference website, including instructions on how to book delegate tickets and find out more about exhibit opportunities. Both 2016 and 2017 shows were extremely popular with attendees and so we advise registering your interest early.

In silicon photonics, optical components are fabricated onto conventional silicon wafers alongside electronic microcircuitry. These components can include sophisticated lasers, elaborate waveguide structures, detectors, modulators, delay and multiplexing/demultiplexing structures, and many other integrated photonic components. The chips are then integrated into packages that can include other chips and active elements plus lenses and other miniaturized optics, plus of course optical fibres and fibre arrays and electrical I/O. 

Precision alignment of the photonic elements is necessary for this, and for testing the chips to ensure their functionality before the packaging process even begins. Both testing and packaging requires that optical inputs and outputs be coupled to sources and detectors, and all elements must be tested and aligned to each other throughout the optical path to ensure efficient coupling. For today’s SiP devices, transverse alignment tolerances of much less than 50 nm are increasingly common. For devices with multi-channel photonic inputs or outputs, additional precision optimization around Theta-Z is needed for efficient coupling of all inputs and outputs of the arrays. It is often the case that Theta-X and Theta-Y orientations also need to be optimized.

These specialized, highly precise and often multi-degree-of-freedom positioning tasks — repeated multiple times from wafer to final package — mean the engineer is confronted with vexing geometric effects. For example, adjusting an angle will translate a coupling in X and Y if the centre of rotation is not perfectly positioned (meaning, practically always). In addition, the increasingly commonplace short SiP waveguides exhibit steering effects so that optimization on the input side causes a shift on the output side. That means that the overall best alignment becomes a "moving target". In the past, an iterative approach was therefore necessary to address this and achieve an acceptable, consensus alignment. Above all, this type of process was very time-consuming.

PI has met these challenges and integrated completely new, firmware-based algorithms into its industrial fibre alignment subsystems for fast multi-channel/multi-degree-of-freedom photonics alignment.  Designed for integration into production tooling ranging from wafer probing through chip-test to final package, these automated subsystems perform multiple linear and angular digital gradient search optimizations simultaneously.   

The resulting parallelism can eliminate the iterative approach formerly required, resulting in process throughput improvements that can exceed two orders of magnitude.  Clearly, the impact on production economics and competitiveness is profound. Moreover, the inherent parallelism of this closed-loop, all-digital technology means the overall alignment time in multi-channel/multi-degree-of-freedom applications is only weakly related to the number of individual alignments being performed.  For example, in the case of wafer probing operations, it is typical that waveguide I/O coupling is optimized in less than 500msec regardless of the number of inputs or outputs to the SiP chip. Similarly impressive throughput is seen throughout subsequent chip-test and packaging processes.

Because different devices and production applications present different requirements, there are several fast positioning system variants for alignment of single-channel or array devices, with or without angular optimization requirements. All systems are based on very stiff set-ups. If angular optimization is not required, the most popular configuration begins with three stacked precision linear motion stages. 

These provide 25mm of travel for coarse positioning and first-light seek.  Mounted at the end of this long-travel stage stack, a fast NanoCube positioner provides fast areal scanning motion for mode localization, and fast transverse and Z gradient search capability for dynamic compensation of drift effects, all with nanoscale resolution. Flexure guides and all-ceramic insulated piezo actuators guarantee a long lifetime in round-the-clock industrial deployment. Position sensors on all drives provide microsecond-scale position determinacy, enabling soft limit capability that can safeguard the machinery and costly devices like fully-patterned wafers.

If angular optimization is required, for example when aligning photonic array devices, a parallel-kinematic hexapod is used to provide long-travel, six-degree-of-freedom positioning and first-light seek, plus fast automated angular optimization. These hexapods provide a freely definable coordinate system and pivot point, allowing rotations to be performed about optical sweet-spots such as beam waists and physical channel centres. The NanoCube positioner is again deployed for fine transverse and Z alignment plus dynamic compensation of drift effects and residual geometric errors in parallel with angular optimization processes.

The combination of hexapod plus Nanocube provides the foundation for the groundbreaking parallel alignment capabilities of PI’s FMPA systems. For example, to align a linear or 2-D fibre array, the NanoCube performs a transverse gradient search with tracking to keep the first channel of the array aligned, while the hexapod performs a parallel theta-Z gradient search on the Nth channel of the array. The hexapod’s freely definable center of rotation has already made sure that the optical axis of the first array element can be positioned near to the first channel’s optical axis; any small remaining geometrical errors are then compensated by the tracking of the NanoCube. The parallelism of the overall alignment means the entire process is 10-100X faster than previously possible with previous-generation iterative approaches.

These systems are offered in standard configurations for single- or double-sided alignment tasks, with or without angular optimization.  Furthermore, the modular architecture allows additional alignment robotics to be provisioned to meet virtually any applications need.  The scope of delivery includes a high-performance digital controller (E-712), the firmware routines with the algorithms for fast positioning and an extensive software package, which covers all application aspects, starting with easy start-up to the convenient control of the systems using graphical user interfaces to fast and manageable integration into external programs. 

For systems requiring angular optimization, the hexapod controller (C-887) implements the same scanning and alignment algorithms.  These include built-in sinusoidal and spiral areal scans with automatic modeling for accurate peak localization even with fast sparse scans, automatic centroid-determination in top-hat and other challenging couplings, and the unique parallel gradient search processes for one-step optimization across inputs, outputs and degrees of freedom.

Together, these novel subsystems help ensure cost-effective testing and manufacturing of today’s silicon photonics devices— and tomorrow’s. 

Imec is investing heavily in this technology - last year, our researchers demonstrated a unique germanium-silicon modulator with a 56Gb/s bandwidth. This modulator enables compact, efficient optical links to be made for server-to-server communication in data centres. Recently, this modulator was developed further and used in a 16-channel 896Gb/s prototype transceiver, integrated in a photonic chip measuring just 1.5mm². What’s more, the modulation speed per channel can be increased further to 100Gb/s, as demonstrated in conjunction with Ghent University (UGent). 

Every day we produce massive quantities of data. You only have to think of e-mails, attachments, mobile messaging, pictures and videos posted on social media, activity logged by fitness trackers, and the list goes on, to see why. At the same time, the quantity of data processed by companies such as Google, Facebook, Microsoft and Amazon is growing exponentially. And then there’s the IoT where all kinds of things will be equipped with sensors taking measurements and communicating with one another. All of which means even more data. 

Typically, this data is stored and processed in the cloud - in other words, in a data centre. Data centres are where thousands of servers sit quietly in racks, all communicating with one another via a complex network of optical fibres. To meet growing demand, this network also has to be upscaled exponentially. The roadmap for this server-to-server communication is extremely ambitious: since 2016, the most advanced cloud data centres have installed optical links and transceivers with a capacity of 100Gb/s; By 2019 an upgrade to 400Gb/s is expected and by 2022, this will be upscaled further to 1.6 Tb/s. 

Cloud data centres can be very large facilities and the optical links they contain need to have a range of at least 500 meters. The optical links for cloud data centres also need to be manufactured in much greater volume and at substantially lower costs than many traditional telecoms solutions. 

Silicon photonics is an interesting technology for integrating the essential building blocks required for an optical link in a single chip. The big benefit of this technology is that the optical components can be produced using the same advanced devices with which microchips are also made. This makes silicon photonics components relatively cheap. Better still, they make a high integration density possible, while consuming less energy and guaranteeing a high yield. 

Imec has developed a platform for silicon photonics for high-speed optical links for data, telecom and sensing applications (voor hogesnelheid optische links voor data- telecom- en sensor-toepassingen). It uses 200mm and 300mm SOI wafers (silicon-on-isolator wafers) as its substrate. And the manufacturing process is based on a modified 130nm CMOS flow, expanded by 193nm lithography to produce the waveguides, using germanium for the photodetectors. An additional oxide/poly-silicon stack provides greater freedom when designing the optical components. This stack is used to integrate passive components such as grating couplers for optical fibres, waveguides and multiplexer filters. 

With the integration platform, both passive and active components (such as opto-electronic modulators, heaters and germanium-on-silicon photodetectors) can be integrated. Electronic circuits (such as drivers and transimpedance amplifiers -- or TIAs) can be made on a separate chip and assembled with the silicon-photonics circuit using flip-chip techniques (packaged into one system). In these circuits, the driver converts a standard CMOS-bit signal into an electrical current that is compatible with the optical chip, while the TIA amplifies the photo-electric current into a standard CMOS-bit signal.

Working in conjunction with researchers from UGent, imec has presented a real-time single-channel 100Gb/s non-return-to-zero on-off-keying optical link in silicon photonics. The same ultra-compact GeSi modulator was used for this configuration. It was combined with a transceiver chip designed by UGent in SiGe BiCMOS technology. The signal transfer enabled by the optical link was tested over a standard single-mode optical fibre (SSMF) 500 meters long and a dispersion-shifted optical fibre (DSF) 2 km in length. The transmission was successful without any complex digital signal processing (DSP) being required. And this result shows that silicon photonics is the right technology for producing compact, low-power transceivers with scalable capacity for future server-to-server connections.

The GeSi modulator used in the two demonstrators above is available for companies and research groups, as are other components from imec’s silicon photonics platform. This can be facilitated via the imec-ePIXfab SiPhotonics:iSiPP50G service, which is part of the Europractice silicon photonics multi-project wafer (MPW) service. 

Lab on chip (LOC) devices combine certain laboratory processes to achieve an automation of a sensing protocol. The LOCs are employed at the Point of Care (POC), which is highly application dependent and could be at a hospital, a doctor's office or in a remote area. Typically, LOCs are developed to allow diagnostics outside a central laboratory by the automation of manual handling steps like the mixing, filtering and heating of a number of liquids. The LOCs focused on in this article are composed out of a biosensor, a certain fluidic circuit, reagents and actuation mechanisms to allow the controlled manipulation of the liquids.

The ideal LOC for biological applications uses a biosensor that is small, but delivers a broad range of information with a high sensitivity, selectivity and robustness. The lock-and-key principle uses capture molecules, which selectively bind to the target. If there is an array of different capture molecules on a substrate, the system is called multiplex. A typical sensor of that type uses a substrate (e.g. glass, polymer etc.), which is bio-functionalized with, for example, 82 spots in the form of nine rows and nine columns shaped as circular spots with a diameter of, for example, 200 µm and a spacing of 500 µm. The sensor is positioned on a channel with the active area of the sensor facing towards the liquid within the channel. Selectivity and dimensions of the array are mainly affected by the biochemical procedure (so called assay) and the technological capabilities to functionalize the sensing areas. Sensitivity and robustness are mainly influenced by the sensor system underneath the biological layer.

Biosensors for mobile applications feature the capability of being potentially highly miniaturize-able. The used liquids are typically in the microliter range, despite very rare targets within a sample liquid (e.g. blood, saliva), where pre-concentration steps need to be applied. Once the sensor has been implemented, the remaining elements of a disposable LOC, the microfluidic channel network and the actuation mechanisms, become the focus of the miniaturization. 

The microfluidic channel network can be produced out of polymer materials. Metals and glass are not preferable since they are too expensive for disposable applications [2]. The most famous material for microfluidic prototypes, when looking at research papers, is the silicone Polydimethylsiloxane (PDMS). It is easy to use for the casting of microfluidic structures but also comes along with some drawbacks, which excludes it from close-to-market, low-cost applications [3]. Therefore, it is recommended to transfer microfluidic networks to well-known, state of the art technologies like standard injection molding as soon as possible.

When it comes to additional actuation functionality, the same boundary conditions with respect to the microfluidic channel network have to be taken into account. Among other technologies, the electrochemical actuation technology developed by Fraunhofer ENAS shows unique features with respect to size, cost and integrability [4], [5], [6], [7]. 

This article illustrates how photonic integrated circuits can be transformed into a LOC system by considering the project Photonic Biosensor for Space Application (PBSA) funded by the EU, which focuses on the integration of microfluidic actuation, a photonic sensor and the bio-functionalization as well as the biochemical protocol into one LOC. The whole system was meant to be included in a small enclosure, which carries all necessary supporting infrastructure to drive the LOC. The target application was to analyse extraterrestrial soil for traces of life in terms of biological targets. State of the art analysis uses spectroscopic approaches, which list up elements present in the sample. The detection of bacteria or enzymes relies on wet chemistry approaches, which aren't available for space yet [8–10]. 

Modifying the PIC for biosensing

The PIC has an outer dimension of 9.5mm square and was developed by DAS Photonics, Spain. It combines a light input port with optical ring resonators and an optoelectrical interface (Figure 1a).  Figure 1b shows the PIC's layout and marks certain important areas, which will be described further on. Starting at the bottom left corner of the scheme, the light enters the chip through optical couplers (“Light Input Port with Grating Couplers”) and is guided through waveguides (“Waveguide”) to the sensor areas of the chip. To achieve a multiplex sensor, the signals are split into an array of multiple sensor elements called optical ring resonators (the blue underlain area on the right; “Optical Ring Resonator”). Outgoing optical signals are transformed to electrical signals (“Light-To-Electrical Signal Interface”) by integrated photodiodes (“Photodiodes”). The collected electrical signals are transported through bonded wires (“Electrical Output Port”) to a signal amplification and interpretation electronic within the electronic management system. 

The reagents and the sample get pumped over the surface of the PIC along with the "Sample Flow Channel" (figure 1b). Around this area, a sealing will ensure the leakage free progress. The necessity of a sealing demands a certain increase in the dimension of the PIC. The optical ring resonators are not covered with an oxide and can interact with the stream of chemicals within the sample flow channels (figure 1 c3). A signal change can be observed during any change of the media. To achieve a selectivity for a specific biosensing application, the ring resonators need a surface treatment and bio-functionalization. 

Bio-functionalization and biochemical procedure 

The bio-functionalization is a procedure to bind antibodies to the optical ring resonators which was done by the Centro de Astrobiología (INTA-CSIC), Spain [11]. The antibodies are the major element for the afore mentioned selective key-lock-binding principle of a certain target within a sample. Figure 2a shows a 3D illustration of an optical ring resonator CAD model. On top of the blue substrate, the optical guiding structures are shown. The reddish cuboid represents the liquid flushing over the surface. The Y-shaped, yellow objects are the antibodies, which are commonly represented in that form. The antibodies are bound to the surface by printing small droplets with a certain antibody concentration on the surface after the surface was primed by a sequence of chemicals. The optical ring resonators only sense changes happening within their evanescent field. An immobilization of antibodies outside this field would lead to not detectable binding events and to a certain decrease in the concentration of the target within the sample. In conclusion, the sensor will have a lower sensitivity in comparison to the potential sensitivity the PIC is capable to deliver [12].

The biochemical procedure on one fully functionalized ring resonator is as follows. First, a buffer solution gets pumped over the ring. The shift in the resonance frequency of the ring is transformed into the first signal level increase as shown in figure 2-b1. The sample solution typically comes with a variety of different, potential biological targets. As soon as this mixture arrives at the sensing elements, a diffusion based, selective binding of the target within the sample to the immobilized antibody occurs (figure 2-b2). After the sample volume got pumped over the surface of the sensor, a washing step follows, which washes away all compartments, which have not bound to the antibodies (figure 2-b3). If the sensitivity of the sensing element and the amount of target within the sample, which has bound to the antibodies leads to a sufficient amount of signal increase, additional steps are not necessary. The biochemical assay would, in this case, be called label-free. To enhance the signal of the sensor, additional compartments can be bound to the target on the rings. This could be a secondary antibody (figure 2-b4 and 5) and for example an additional nanoparticle (figure 2-b6 and 7).

Assay miniaturization and sensor integration

With the functionalized sensor, the detection system is ready, but relies on personnel to manually pipette the liquids on top of the surface of the sensor, step by step. A microfluidic cartridge can work as a passive interface for the liquids from an external reservoir to the sensor as depicted in figure 3. The fluidic cartridge is composed out of a milled polymer chip, which guides the liquid from certain tube ports to the PIC's sensitive area (figure 3a). The connection is made by clamping to allow a reuse of the PIC for this prototyping stage. The system relies on external pumping actuation with external reagent reservoirs. The PIC is placed on a PCB and connected to it by bond wires. An optical fibre connects to the optical couplers on the PIC and an amplifier circuit collects the electrical signals to track the resonance shift of the different optical ring resonators (figure 3b). Such a setup can be used to validate the interface of the fluidics to the sensor and to experiment with volumes and flow characteristics of the single process steps for assay transfer reasons. 

The necessity of external actuation and reservoirs decreases (besides other drawbacks) the user friendliness. Systems which want to be used by customers outside a laboratory, especially used by medically unskilled personnel, need to integrate actuation mechanisms. Eventually, the user only needs to put disposable cartridges into a reader after loading the cartridge with the sample. No additional handling besides the plug & play -nature of the cartridge exchange should be carried out. The microfluidic cartridge technology developed at Fraunhofer ENAS, Germany is capable of storing different liquids in certain cavities within a polymer part. The twist behind this technology is the integrated actuation principle, which is highly miniaturized to allow each individual liquid reservoir to be equipped with its own pump. 

Preliminary to the application specific cartridge design, an intermediate prototype was developed. This prototype is composed out of the photonic sensor together with a standard cartridge called flex.flow microscope slide, which is commercially available and was kindly provided by BiFlow Systems GmbH, Germany (figure 4a). The flex.flow cartridge relies on the afore mentioned actuation technology, is not application specific and can be purchased as an evaluation kit. The cartridge does not include a biosensor but allows developers of biosensors to combine their sensor with a cartridge, which actually makes it a Lab-on-Chip device already. The port for the sensor is simply designed as two holes on top of the cartridge (the holes are positioned within the red frame of figure 4a). The first hole allows liquids to get pumped from internal reservoirs to the sensor. The second hole guides the liquid from the sensor to a waste reservoir. The holes on the cartridge have a certain distance, which is bigger than the PIC itself. To bridge this, the sensor was embedded into a stack of polymer sheets. This stack guides the liquids over the surface of the sensor (figure 4b; blue arrows). To verify the proper functioning, an experiment was done without a bio-functionalization of the PIC. Four different concentrations of the buffer solution PBS were used. Each liquid was placed into a different reservoir of the combined cartridge-sensor prototype (figure 4c). The evaluation kit uses a small power source and a plugging port (comparable to a USB port) to drive each individual pump on the disposable cartridge. The cartridge was programmed to pump the different concentration over the surface of the PIC one after another and the response of the sensor was recorded (figure 4d). The sensor was able to track the different concentrations.

The cartridge used here is a standard system from BiFlow Systems. It is versatile enough to be combined with different sensor systems and to drive an amount of liquids. Since it is not specially designed for one purpose, volumes and assay steps are limited. Each assay is different and therefore, a miniaturization and transfer to the microfluidic cartridge has to be done. Both prototypes were used to validate all necessary assay steps and the bio-functionalization. Concentrations, flow velocities and volumes were analysed and the results were used for a specialized cartridge design.

The design allocates each assay step (or rather each liquid getting pumped) to an individual reservoir and pump. Elements were included that trace back to the experiences made with the integration of the PIC into the prototypes. Further on, the cartridge was extended by fluidic connection and a valving system that allows an external robot to fill samples into the cartridge in an automized or remote control manner. As a boundary condition of the projects target mission, enough functions were integrated to allow storage and processing of two consecutive assays to be performed (figure 5).

PBSA system overview

All necessary components for the control of the cartridge and the sensor were combined into a single enclosure. The system level design was made by Evoleo Technologies, who are expert payload developers and satellite integrators for space applications. Radiation hardened and mechanically durable components and designs were used to fulfil the project’s requirements for the demonstrator. The top three layers provide typical, necessary support components as well as an electronic control system including an operating system and the control algorithms for the microfluidic cartridges (figure 6a and b). The two layers on the bottom provide the PIC with an optical input signal and a read-out of the sensor signal. The laser component has an additional mechanical guiding structure on top to allow a proper integration of the fibre into the enclosure and a proper alignment to the PIC. The middle layer of the system is the microfluidic cartridge in combination with the PIC known from figure 5. 

The operability of the system was shown within the project. The different reagents for the assay were preloaded into the cartridge, the cartridge was closed and in combination with the PIC included into the enclosure. After the sample was loaded into the system through the tube connectors on the bottom of the enclosure, the automatic detection procedure was started. The liquids got pumped over the sensor and the sensor's response was recorded. The target within the sample was measured selectively but with a low sensitivity in comparison with prior experiments in the laboratory. Some of the experiments were already published open source [13].

Summing up

Lab on chip devices can be composed of a varying set of actuation and sensor principles. The focus of this article was to describe how a photonic integrated circuit can become a biosensor and what has to be done beyond this to develop a Lab on Chip device. The application scenario was the project PBSA, where a system for biochemical analysis on the basis of wet chemistry was developed that is potentially capable of performing analysis in a harsh and remote area on earth or during planetary exploration missions. 

Applications, where it is necessary to specifically bind a target to achieve a significant result, rely on disposable, sensing system, as long as a total (biochemical) reconditioning of the system can't be guaranteed. This leads to the question of exchangeability of the consumed parts after each measurement. The PBSA system was based on the mission scenario, where the instrument is a payload of an autonomous rover. For the project it was stated to be easier to equip the rover with multiple instruments instead of a manipulator for the exchange, aligning and fixation of the LoC within the instrument. At this point, there is no final answer to the question of exchangeability for remote applications of LoCs. But for typical human or veterinary diagnostic applications, exchangeability has to be implemented into every product development.

Future thoughts

In general, the applicability of PIC's for Lab on chip systems can be increased especially by the optimization of three parameters of the sensor: One-time usage, connectivity and necessary supporting systems. One-time usage is high with a sensor that uses cheap materials. A material mix analysis done by Yole Développement within the study on "Microfluidic Applications 2015" ([2], pg. 42) showed that the disposable segments (Clinical & Veterinary diagnostics, Point of Care diagnostics) use silicon in only 1 to 6 percent of the application cases but polymers in between 77 to 87 percent. Silicon based biosensors would increase their importance for the Point of Care market by transferring their technologies to polymer materials. In 2013 Hu et al. published a paper on the fabrication of photonic structures on flexible foils [14]. Photonic sensors based on such technologies could be manufactured in a roll to roll manner leading to an enormous scaling effect on production cost, which would also increase the one-time usability and the applicability of Lab on a Chip systems. The alignment of a fibre or a fibre array to the PIC is an element, which should be simplified for an easy exchange of the LoC after use. A plug & play solution would increase the user friendliness of the system. The necessity of a laser as an infrastructure component for the sensor reduces the mobility of a PIC-based Lab on Chip instrument. Especially looking at competing sensor technologies like electrochemical or fluorescence based technologies, PICs have to argue about their unique selling proposition.

PICs have unique features, but there are competing technologies, which are also suitable for biosensing applications. At this point, there is no killer application or technology which will exclude PICs as biosensor principles for point of care tests. The first one, who occupies important niches or applications, will make it harder for other players to enter the market. But we also see a broad range of different providers of point of care tests working in the same application field nowadays, especially with rapid home glucose test systems. This is an application where the commercial spread is extremely low because of the low price of each test strip. It is surprising that there doesn't seem to be a displacement but a growing market here. The applications, which will use an active fluid control on the disposable system, will have higher prices and need to provide solutions for more complex, diagnostic protocols, where lateral flow strips can't be applied. For example - DNA analysis, rare target samples, bio markers with high similarity, assays with the necessity of sample preparation and all multi-marker and quantitative tests. These boundary conditions can be provided by some technology combinations including PICs. The future will show, which one will be the first on the market. But we are sure that we will have a highly diversified market of LOC's with different technology combinations existing in parallel in the future. 

For this reason, the pharmaceutical industry is heavily controlled to ensure the composition of drugs administered to patients. Related chemical processes require strict quality checks to assess the purity, reproducibility, homogeneity, and warn of the presence of contaminants, among other factors. Current evaluation techniques include sampling, transportation to a specialized laboratory, purification of the sample (which is done by skilled personnel), and the final detection using bulky and expensive equipment. Since these actions are time-consuming and costly, and usually the sample is destroyed during the analysis, only a small fraction of the final drug can be analysed.  In addition, these analytical tools are bulky and requires optimization and calibration steps, which creates a hurdle for in-line monitoring.  

In many cases, batch pharmaceutical manufacturing (where all the materials are charged before the start of processing and discharged at the end of processing) is now replaced by cleaner, flexible and more efficient continuous manufacturing, which can avoid off-line delays. In a continuous manufacturing process, material is simultaneously charged and discharged from the process. 

Continuous manufacturing has some advantages when comparing with batch manufacturing such as no manual handling is required, increased safety, shorter processing times, more flexible operation and smaller ecological footprint. However, how to warranty that the product has a uniform content and quality within specified limits? Clearly, the answer is related to the implementation of in-line monitoring detection tools in the manufacturing process.

Mid-IR platform

Mid-Infrared (Mid-IR) technology is based on the strong interaction of light with molecular vibrations. Spectroscopic sensing in the Mid-IR wavelength band (3-12 µm) is a powerful analytical tool since the chemicals exhibit their fingerprint region, intense adsorptions that allow unambiguous identifications and quantifications of molecules.  

A Mid-IR sensor consists of: i) a laser source, usually Interband Cascade Lasers (ICLs) and Quantum Cascade Lasers (QCL), ii) the passive components (PICs) or free-optics, and iii) a detector (type-II InAs/GaSb superlattice (T2SL), InAsSb and Quantum Cascade Detectors (QCD)). Packaging of the final devices includes the integration of the photonics components and the electronics on the same platform which reduce the size of the sensing system (see image below). Other advantages are a high sensitivity and selectivity, which allows unattended, direct and fast detection of the sample without the requirement of any pretreatment - fundamental requisites to integrate these devices into manufacturing lines.

The increasing attention of the scientific community to Mid-IR sensing has driven several studies demonstrating the enormous potential of the Mid-IR technology (see Ref [1] as an example). Regarding new devices for quality control in pharma, it is helpful to highlight the work recently presented by Li et al. in which they develop a Mid-IR imaging system enabling mapping both active pharmaceutical ingredients and excipients of a drug tablet [2]. 

Contact us on the details below to talk about your editorial, advertorial or advertising needs.

E: sukhi.bhadal@angelbc.com

W: www.picmagazine.net

T: +44 24 7671 8970